Machine Learning Basics
What Is Machine Learning – Machine Learning Basics
The simplest answer to this question would be “the future.” The process of machine learning will be a fundamental part of how the technology or even the world continues to evolve and advance. In order to understand the concept as a whole, it is necessary to understand machine learning basics.
So, what is machine learning? The best way to approach this is with an example. If you use social media, you’ll notice that networks like Facebook can automatically recognize faces in photographs or videos. That allows you to tag your friends and your family or even tag someone in a funny meme. Let’s be honest, you’ve probably never thought about how this works or why but you should. Understanding this relatively simple software capability will help you grasp the more complex aspects of machine learning.
Facebook’s ability to recognize faces in photos is automatic. There’s no one on a computer somewhere, staring endlessly at the selfies that you took while you were drunk last night. Well, there might be, who knows what Facebook employees are up to these days. But they’re definitely not coding each one for tagging, they don’t need to. It’s all handled by a piece of software.
A single piece of software has learned what a human face looks like in a photo. In doing so, it can make sure that only sections of the photo with a face are coded for tagging. This is a basic way that machine learning is used today, and some insight into its future use (artificial intelligence).
A Machine Learning Example:
Still confused? Okay, let’s say you we want computer to perform a task. How do we accomplish this? Simple, we create a program for that task. So, for instance, you might want a computer to say ‘hello.’ If you’re not a software developer, you might be surprised to learn that’s actually not difficult at all. A simple program or piece of code will enable a computer to do this.
Let’s go further then and say we want a computer to respond to different commands. So, if you ask a computer a question, you want it to respond. More than that, you want the computer to respond correctly, providing an accurate answer. This is what machine learning allows. With machine learning, we don’t just create programs for automation. Instead, we enable programs to create programs. In essence, we automate, automation.
That ‘learning’ in the title isn’t a coincidence either. Think back to the software that sets up photos for tagging on social media. To do that, the software must be able to learn the traits and features of a face so that it can identify one. This should help you understand what machine learning is, but we’ve only scratched the service of what it’s capable of.
What Can Machine Learning Be Used For?
How much time do you have? Right now, any technological process that you can think of probably involves machine learning. How does junk mail get sorted into your junk box? Machine learning – obviously they’re still perfecting that one. How does a search engine know which web pages to show you in the SERPs? Recently it was revealed that Google has started using – you guessed it – machine learning, to do this. Autonomous cars, marketing analysis, and computational biology all use some form of machine learning. Some processes that we take for granted today are only possible due to this advance in technology.
Why Is Machine Learning So Important?
Again, the best way to explore this is to look at a real-life example. This time, let’s think about drugs. Medical drugs are created by making different compounds, mixing chemicals together and making solutions that could do anything from getting rid of your headache to curing a fatal disease. Unfortunately, regardless of its use, it takes years to find a drug that works. Once you decide on a compound and a formula it then needs to go through various forms of testing. We’re talking about animal testing, human trials and after all that, about five years later, it may get medically approved for the market.
Obviously, it is important to make sure that a drug a) works and b) is safe before it’s sold on the market or used in hospitals. But that still doesn’t change the fact that it’s an incredibly inefficient process. With machine learning, this can change. Theoretically, using machine learning, you can generate a program that could learn to siphon through millions of possible compounds to find the right one. This drug could cure a disease like cancer, and it gets better. Rather than testing on animals or humans, the software will create simulations or individuals, test the simulated drug and find the outcome. Wait, because we haven’t even told you the best part.
It will do all that in minutes. This should give you an idea of why Bill Gates famously said that a significant breakthrough in ML could be worth ten Microsofts. Medicine is just one example. Theoretically, machine learning will change how we think about everything. Cryptocurrency for instance. Right now, cryptocurrency needs miners to operate. Machine learning could eliminate that need completely.
So, we’ve looked it what it is and why it’s interesting. But how does it work?
Understanding How Machine Learning Actually Works
Machine learning works with algorithms. Regardless of the algorithm, there are three basic components that are always present.
First, there’s the representation or the way in which the algorithm is able to represent the knowledge. This could be a decision with different trees for each possibility or outcome.
Then, there’s evaluation which might be a likelihood or a cost margin or level of accuracy. Doing this the candidate programs can be evaluated so the right one can be chosen.
Finally, there’s the optimization where the right candidate program is optimized and generated. Yes, I know, it’s become a tad more technical, but an easy way to think about it is like this.
For typical programming, data and the program and the input into a computer. This provides an output.
With machine learning, data and a potential output are the input into the computer. This then creates a program.
There are two main ways this can happen. Since this article is about Machine Learning basics, I’ll keep it relatively simple.
Different Types Of Machine Learning
Supervised Learning – Machine Learning Basics
Supervised learning provides the program with the output paired with their correct inputs. In doing so, the program can then learn to recognize patterns however this can be impacted by human bias. For instance, you might teach a program to recognize what fish are. Your examples of data will impact what is and isn’t recognized as a fish by the program.
The goal then is to provide the outputs and inputs and then get the program to map the functions to such a precise and accurate level that it will be able to complete the process automatically afterwards. In the case of the fish, the generated program will be able to distinguish what is and isn’t a fish based on the inputs and correct outputs provided.
Unsupervised Learning – Machine Learning Basics
Unsupervised learning is more complex. In this case, the program doesn’t receive the correct outputs, only the inputs. As such, the role of the program is to explore the data and find the important variables, possibly recognizing patterns. The earlier example of one day being able to find the right formula for medicines that work would be an example of what this type of ML could lead to.
I hope this helps you understand some Machine Learning basics, what it is, what it could become and why it’s so important in virtually every industry today.
- Machine Learning Basics - April 16, 2018
- Vital Information About Cryptocurrency Wallets - February 12, 2018
- The Complete Guide to Altcoin: Part II - February 5, 2018
- How to Invest in Bitcoin: A Complete Guide - January 29, 2018
- Blockchain Explained - January 22, 2018
- The Basics of PPC Management - January 18, 2018
- Altcoins – What are they, and why are they useful? - January 15, 2018
- Beginner’s Guide to Backlinks from a Link Building Expert - January 11, 2018
- Ultimate Bitcoin Guide: How Does Bitcoin Work - January 8, 2018
- How SEO Works: What Every Business Owner Should Know - January 4, 2018